1. Field of the Invention
The present invention relates to an image processing device and an image processing method, and particularly, to an image processing device and an image processing method able to embed information into or extract information from an image.
2. Description of the Related Art
In recent years, along with improvement of image processing techniques and image forming techniques, one can even faithfully duplicate notes (paper money) or securities with a digital color copier with little difference between the duplicated ones and the original ones. For this reason, for special documents like the notes or the securities, it is necessary to take measures to prevent such special documents from being illegally duplicated or from being correctly duplicated.
Additionally, for example, in a company, instead of the special documents like the notes or the securities, even for general documents, since there are many confidential documents, it is necessary to control duplication of such confidential documents from the point of view of confidentiality. That is, it is necessary to take measures to prevent such confidential documents from being illegally duplicated or from being correctly duplicated.
Due to this, in the related art, many studies have been made regarding restricting duplication of the special documents and confidential documents. For example, Japanese Laid Open Patent Application No. 2004-274092 (hereinafter, referred to as “reference 1”) discloses a technique in which output is prevented if a predetermined dot pattern is detected from an image read by a scanner. Therefore, if the predetermined dot pattern is embedded in a special document or a confidential document, duplication of which is not allowed, it is possible to effectively prevent the document from being reproduced.
Although the technique disclosed in reference 1 is able to effectively prevent a confidential document from being illegally duplicated, this technique can deal with only a small amount of the information embedded in a special document, specifically, only one bit indicating whether the target document is a confidential document. However, in order to realize information security functions of high flexibility, for example, the embedded information is combined with user authentication, and duplication permission is switched depending on the position of the user, so that it is necessary that the embedded information be of at least a few bits.
In addition to preventing output during duplication, when using a copier without the function of detecting the output prevention data, like the dot pattern, when a copy of the document obtained with the copier outflows, it is also required to embed tracking information into the document to determine from where the copy outflows. In this case, it is desired that the amount of the embedded information be about 100 bits or more.
To attain these objects, the inventors of the current invention proposed an information embedding method involving background dot patterns and enabling extraction of about 100 bits of embedded information. For example, this technique is disclosed in Japanese Laid Open Patent Application No. 2006-287902 (hereinafter, referred to as “reference 2”).
There are many other methods enabling extraction of about 100 bits of embedded information, and these methods have advantages and disadvantages. For example, in the method utilizing a background dot pattern, since the dot pattern is repeatedly embedded in the background of the document, the dot pattern more or less influences reading of the document, but this method has good hiding properties, in other words, the dot pattern cannot be hidden easily during duplication.
On the other hand, methods involving embedding of a code image in a specified region, like a common bar code and a two-dimensional bar code, have good versatility, but can be easily hidden during duplication when used for preventing un-authorized duplication.
Further, when text images are used, by changing character intervals or changing the character shape, it is possible to embed information with the embedded information being hardly perceivable by human beings.
For example, T. Amano and Y. Hirayama, “A method for embedding digital watermarks in page descriptions”, Information Processing Society of Japan (IMPS) SIG Technical Report, Vol. 98, No. 84, Sep. 17, 1998, pp 45-50 (hereinafter, referred to as “reference 3”) discloses the method of changing character intervals to embed information.
H. Tulia and M. Mesuji, “Digital watermarks in lettering Images by Using Character Shape”, The Transactions of the Institute of Electronics, Information and Communication Engineers. D-II, Vol. J82-D-II, No. 11, Nov. 25, 1999, pp 2175-2177 (hereinafter, referred to as “reference 4”) discloses the method of changing the character shape to embed information.
Therefore, it is desirable that depending on applications, one or more information embedding methods be appropriately selected among plural kinds of information embedding methods to embed information, thus improving convenience.
However, if it is allowed to arbitrarily select a method of embedding multi-bit information from plural information embedding methods, when extracting the embedded information, it is impossible to determine by which method the information is embedded. Thus, it is necessary to try extraction of the embedded information with all possible information embedding methods, and this degrades performance of information extraction.
In addition, the information extraction process is performed even for general documents in which multi-bit information is not embedded, and this degrades performance of the duplication process for the general documents which do not need duplication control.
Generally, when the amount of information embedded in a specified area of a document increases, extraction of the embedded information requires a memory of a large capacity and a great deal of processing. Due to this, when making copies with a copier, it is difficult to extract information embedded by a multi-bit information embedding method in real time, that is, scanning an image with a line sensor in units of lines, while performing the information extraction processing in parallel. Therefore, in order to extract multi-bit information when making copies, it is required that the output process be stopped before the whole image, or an amount of image information sufficient for information extraction, is loaded in a frame memory, and this causes non-negligible degradation of the performance of the above duplication process.